Scoping review to understand the potential for public health impacts of transitioning to lower carbon emission technologies and policies

Background: The transformation of the global energy sector from fossil-based fuels to low/non-carbon fuels will reduce environmental pollutant load, which in turn will benefit human health. However, with upscaling of emerging renewable technologies and energy sources, it is important to identify the potential for unintended health impacts, and to understand where the knowledge gaps lie with respect to health. We aimed to identify these gaps by conducting a scoping review. Methods: We conducted a systematic search of Medline, Web of Science, PubMed and EMBASE. We used broad search terms to capture literature associated with energy transitioning to low/non-carbon energy sources or related technologies, combined with terms relevant to measuring or estimating health outcomes/impacts associated with environmental exposures. We included original epidemiological studies, reviews, health impact assessments (HIAs), life cycle assessments (LCAs), and modelling studies that examined health impacts. Results: The search identified 6933 papers of which 81 original research and review papers were included in the review. The majority of studies were based on modelling scenarios. There were few papers reporting empirical epidemiological studies, either observational or interventional. The principal foci of the studies were: alternative energy scenario modelling; biofuels; wind energy; photovoltaic cells; transport; and building energy efficiency. Within those studies the depth and breadth of the health impact research was limited. Conclusions: There is a need to determine the potential for unintended health impacts that may arise from each energy transition scenario, as an adjunct to consideration of environmental and social impacts. Conducting LCAs or HIAs associated with current and emerging transitions, technologies, energy interventions, and policy decisions are likely to be the best methods, currently, for determining the potential for health impacts. Such research needs to be multidisciplinary and iterative to keep abreast of developments in new energy technologies, modelling methods and policy shifts in energy transitions.


Introduction
The current global energy sector is based primarily on fossil-based fuels, namely oil, gas and coal. Fossil fuel combustion has short, medium and long-term impacts on health (Smith et al 2013). In the short-to mediumterms, fossil fuel combustion creates air pollution due to the release of particulate matter and gases that have adverse impacts on human respiratory and cardiovascular health (Wu et al 2018) and cognitive health (Clifford et al 2016). Over the longer term, the contribution of fossil fuels to climate change (Intergovernmental Panel on Climate Change (IPCC (2018)) is predicted to increase health risks associated with more frequent extreme considers the various methods used to date, it was not our intention to evaluate the modelling methods as this was beyond the scope of our review and has been reported recently (Chang et al 2017).
The aims of this scoping review were: (1) to synthesise the current evidence available on the health impacts of transitioning from traditional fossilbased fuels to other low/non-carbon forms of energy in high-and middle-income countries; and (2) to identify current gaps in knowledge on the potential for health impact from energy transitions, processes and technologies.

Methods
As the field of energy transitions is multidisciplinary and broad in its inputs and reach, we used the following criteria to structure the review. The inclusion criteria comprised research which reported empirical and modelling studies, reviews, and HIAs that examined the health co-benefits or impacts in relation to energy transition technologies and processes. As we aimed to include all papers related to health we included 'health' as a broad search term as well as specific search terms pertaining to cardiovascular and respiratory disease (table 1). Given the need to focus the review on health and public health impacts, we excluded research that related to: (1) energy transitions but which did not relate to health outcomes; (2) clinical or public health trials that did not relate to energy transition, energy use or alternative energy production; (3) modelling of climate change policies that did not relate to transitioning from fossil-based energy sources to other low/non-carbon forms of energy. That is, papers which reported analyses of the health co-benefits of improving air quality through general reductions in GHG emissions or changes to policy and planning, as we were mindful of the substantial literature already in existence (Thompson et al 2014, Turnock et al 2016, Partanen et al 2018; (4) changes in building energy use that did not examine impacts on health; (5) energy use for indoor cooking or heating using biomass in low income settings; (6) economic, social, political, technological and temporal impacts on energy transitions which did not related to health outcomes; and (7) studies relating to consumer and societal behaviour, energy security and geo-political science, as we considered these to be outside the scope of the review and are broad, complex disciplinary areas in themselves.
To avoid duplication of included papers we checked the original papers cited in review papers and subsequently excluded them unless they contained relevant findings that were not reported in the reviews.

Search strategy
The literature was searched using bibliographic databases commonly used in the health sciences: Medline, Web of Science, PubMed and EMBASE. No publication date restrictions were imposed but only English language papers were searched and selected.
The search terms used are detailed in table 1. Terms Energy transition energy AND (transition * OR reform * OR future * OR greenhouse gas * OR decarbon * OR fossil fuel * OR alternative OR carbon constrained OR clean OR renewable * OR low carbon OR future * OR solar * OR wind * OR biofuel * OR electricity OR biogas OR ethanol or diesel OR gasoline OR petrol * ) AND Health outcome health AND (co-benefit * OR externalit * OR impact * or effect * OR cardiovascular OR cardiorespiratory OR CVD OR respiratory) Excluding NOT (diet OR biomass OR accident * ) The abstracts of all identified papers were reviewed for initial inclusion, then full papers were read if inclusion criteria were met. Further hand searching of reference lists from included publications was undertaken. We have reported the results of these papers along with discussion of them in the next section.

Results
Our search, conducted on 28 January 2019, initially identified 6933 papers of which 82 papers were included in our final narrative synthesis (figure 1). Overall, few epidemiological studies, either observational or interventional, were found in the peer-reviewed literature. Of these most were related to wind turbine studies and building energy efficiency studies. The majority of papers were based on modelling energy and climate change scenarios. We included 18 review papers which had previously synthesised relevant literature.
The included papers covered a diverse and broad range of key topics associated with the health impacts of energy transitioning from fossil-based fuel to low/no carbon energy sources. To facilitate the summary of findings from the range of papers we grouped them according to themes: modelling alternate energy production scenarios, biofuels, wind energy, photovoltaic cells, alternative forms of transport, and improving building energy efficiency.
The included papers comprised studies that modelled scenarios of alternate energy source/production/use and the impact on health (n = 29), followed by studies on building energy efficiency (n=17), wind turbines/ farms (n=14), transport (n=6), and photovoltaic cells.

Discussion
Modelling alternate energy source/production scenarios The majority of papers included in this review were modelling studies that compared energy source scenarios and estimated health impacts or benefits. The modelling approaches are summarised in table 2 by geographic location. In the USA, the CMAQ chemical transport model combined with a regulatory standard health impact assessment was reported in the US EPA Benefits Mapping and Analysis Program (Akhtar et (Partridge and Gamkhar 2012), were reported. Among these studies, the key health outcomes were assessed using economic modelling, for example, premature mortality or avoidable deaths using the 'value of a statistical life' to monetise the savings, and respiratory and/or cardiac hospitalisations and/or symptoms using the metric 'willingness to pay' to monetise the savings.
The modelling studies of fossil-fuel based energy production that was replaced with renewable energy sources (solar, wind, hydro) resulted in reduced levels of air pollutants (particulate matter with aerodynamic diameter of <2.5mm [PM 2.5 ], <10mm [PM 10 ]; CO 2 ; NO 2 ), with concomitant health benefits being obtained through reduced avoidable deaths and reduced health care costs associated with hospitalisations, emergency department attendances, and loss of time from work/school due to ill-health (Aunan et al 1998, Chen et  . Two modelling studies examined nuclear power and their results suggested that nuclear energy has the potential to reduce premature deaths when compared to fossil-fuel energy (Rosen 2009, Qvist andBrook 2015) due to reduced air pollution emissions even after accounting for potential radiation health risks. One study modelled the impact of replacing ethanol with gasoline in Brazil and reported that the air pollutants emitted by the ethanol were higher than from gasoline and could potentially have more adverse impacts on long-term health outcomes (Scovronick et al 2016). Of these studies, one study used a LCA to model base load power generation using fossil fuel, nuclear, wind, solar and geothermal technologies in Europe. They found that, overall, nuclear and renewable energy and natural gas power generated substantially less human health impacts than hard coal and lignite (fossil-fuels). Fossil fuel combustion and mining (coal, uranium and metal) were reported as generating the highest human health impacts (Treyer et al 2014).
Despite the relative consistency of health benefits related to transitioning from fossil-fuel to lower-carbon fuels and other air pollutants, we were unable to pool the results due to the diversity of modelling methods, assumptions, scenarios, geographic locations and evaluation metrics.

Biofuels
Liquid biofuels are the subject of increasing interest as they represent sources of renewable fuels considered to have the potential for reduced GHG emissions and so their use could assist in mitigating climate change, strengthening energy security and contributing to diversified agricultural economies. Liquid biofuels are usually produced by fermenting sugars derived from plants such as corn grain and sugar cane into ethanol; or by processing oil crops such as canola, soybean or palm oil into biodiesel (Scovronick and Wilkinson 2014). Research is currently exploring the viability of 'second-generation' biofuels which include those produced by the conversion of lignocellulosic (plant cell wall) feedstock residues into bioethanol and other renewable liquid This analysis indicated that emission reductions of aerosols and their precursors under expected USA air quality regulations will lead to significant benefits to human health, yet they will, on net, increase the rate of near-term climate change because reductions in USA emissions of cooling sulphate aerosols will more than offset reductions in warming black carbon aerosols. In the combined scenario where both near-term emission limits were put into place alongside a long-term CO 2 reduction goal, they found opportunities to improve both human health and climate outcomes beyond the outcomes from a single policy. Health benefits varied in order of magnitude with annual benefits ranging from US$75 million for the smallest installation to $690 million for the largest. Benefits attributed to reduced SO 2 followed by reduced CO 2 and then NOx.
Model does not account for: full lifecycle impacts of fuels; seasonal or temporal variation in power plants cycling up and down and the associated emission; particulate matter emissions and other gases and metals.
Air pollution emissions (SO 2 , CO 2 and NOx) modelled from CMAQ.
Variability was associated with facility size, geographic location and simulated year (2012 v 2017). Buonocore et al 2016b; USA Modelling benefits of different energy efficiency and renewable energy choices (wind, solar, peak demand-side management (DSM) and baseload DSM) by displacing the emissions from fossilfuel based power generators across 24 scenarios (in 6 USA cities) using the Electrical Policy Simulation Tool for Electrical Grid Interventions (EPSTEIN) model. Air pollution emissions (SO 2 , CO 2 and NOx) modelled from CMAQ.
Health impacts=value of a statistical life (VSL) of US$7.58 million.
Total health benefits varied by a factor of 37 across the 24 scenarios with central estimates varying from US$5.7 million to US$210 million, with displaced SO 2 from coal generally dominating the benefits. Hence quantifying public health benefits may be site specific and vary by the technology.
Model does not account for: full lifecycle impacts of fuels; seasonal or temporal variation in power plants cycling up and down and the associated emission; particulate matter emissions and other gases and metals. Many generators have air emissions that may be potentially damaging to health. This study found that using backup generators to supply electricity during the periods of peak demand has lower private and social/health costs than a new peaking plant in addition to making electricity supply more reliable and relieving major problems associated with siting new generation and transmission. This analysis uses conservative assumptions throughout that tend to overestimate the health costs.
While uncontrolled diesel ICEs would harm air quality and health, putting controls on these generators and using ultralow sulphur fuel reduces the social costs significantly. Location and maintenance are important considerations.

USA
Comparing the levelised private and social (health) costs of diesel internal combustion engines (ICE) with and without diesel particulate filters (DPF), natural gas ICEs, and microturbines to a new power peaking plant in New York compared with using back-up generators to provide power during periods of peak energy demand.

McCubbin and Sovacool 2013; USA
Modelling natural gas against wind energy in California (Altamont) and Idaho (Sawtooth).
Premature mortality-unit value=US $8.8 million Wind farms likely to avoid the following costs associated with premature mortality: High level of ambiguity in some of the models' inputs: emission rates of air pollutants; location and sources of emissions; estimated number of avian deaths associated with colliding with turbine blades compared with those dying due to climate change. Modelling cogeneration of thermal and electrical energy power using coal and uranium in a range of proportions and supplying differing proportions of the residential, commercial and industrial sectors.
Mortality, morbidity and days of work lost.
Modelling indicated that cogeneration of thermal and electrical energy led to reduced air pollutants and associated reduced health costs. However, the measurement and analysis of the health impacts within the scenario modelling was not clear.
Results validated 20 years after the initial case study was undertaken-annual use of gas, liquefied natural gas, coal and petroleum has increased at higher proportions than predicted in the base modelling. 1990 Method for health impacts not described in the methods.

Partridge and Gamkhar 2012; China
Modelled health co-benefits using 'Damage function methodology' in China: Transition from coal-powered power stations to ones using wind and small-scale hydro projects.
Premature mortality (value of statistical life); incident chronic bronchitis, respiratory and cardiovascular (CV) hospitalisations Preliminary modelling indicated overall reduction in premature mortality, chronic bronchitis and CV and respiratory hospitalisation across all regions of China, but levels varied between regions and types of energy transition.
Limited range of co-benefits were included in the model to fully inform a costbenefit analysis. Authors state their results are subject to considerable uncertainty but provided a preliminary analysis that could inform future research.
Occupational health risks not included.
For wind generation, the co-benefit varied between 2.3% and 9.1% of the additional cost compared to a coal-fired power station in the same region Only estimated health damage related to PM+secondary sulphites and nitrates (excluding ozone and SO2).

Peng et al 2018; China
Modelling coal intensive versus half decarbonized power for electricity production with multiple end-user electrification scenarios in China.

Avoided deaths
Half decarbonized power supply (∼50% coal) for electrification of the transport and/or residential sectors leads to a 14%-16% reduction in carbon emissions compared to BAU, as well as greater air quality Modelling was based on annual total emissions and authors suggested that future models should include a finer temporal analysis. Exposure: PM 2.5 and health co-benefits (55,000-69,000 avoided deaths in China annually) than coal intensive electrification.
Other important air pollutants such as NOx and O 3 . Modelling of the displacement of coal use in power, industry and households with coal-based synthetic natural gas (SNG) in China.
Premature mortality Deploying all SNG to the residential sector can avoid 32,000 (20,000 to 41,000) air pollution-related premature deaths nationwide in 2020.
In contrast, allocating all SNG to the power or industrial sectors barely improves air quality and avoids only 560 (230 to 740) or 3,100 (1,300 to 4,300) premature deaths, respectively.
Due to excess CO 2 emissions from SNG compared to coal, there is a need for an accompanying carbon capture and storage strategy to mitigate effects on climate. SNG produces less air pollutants such as SO2 and PM but has higher CO2 emissions.
These reductions are approximately 10 to 60 times higher than reductions when SNG used in the industrial and power sectors. Model-ECLIPSE_V5a_CLE Base case=2020 Air pollutants: SO2, NOx, PM10, PM2.5

Ramaswami et al 2017; China
Social-Ecological-Infrastructural Systems framework. 637 Chinese citiesdetailed data on energy supply and heat distribution.

Premature deaths
The What-If FYP-Efficiency-plus-Symbiosis compared to Base Case model predicts average premature deaths avoided is 5.6% (25,500 to 57,500) annually. The benefits are highly variable across cities with the mega cities experiencing the greatest reduction in air pollutionrelated premature deaths (28%  Greatest public health cost benefit was associated with PM2.5 reduction. Modelling the displacement of coalfired power stations with solar energy production in China. Premature mortality due to chronic obstructive pulmonary disease (COPD), lung cancer, ischaemic heart disease (IHD), ischemic stroke Balanced_Regional scenario led to greatest health benefit of 10,000 (5000 to 14,000) premature mortalities avoided.
No life cycle assessment of solar panels included in modelling.

Scenarios:
Exposure modelled by ECLIPSE_ v5a_CLE: CO 2 , SO 2 , PM 2.5 , NO 2 Modelling the health benefits of the National Energy Efficiency Improvement and Energy Conservation Program compared to current status; 20% and 30% reduction scenarios.
Health impacts: Mortality, chronic respiratory symptoms, asthma symptom days, lung cancer cases.
Health benefits from reduced air pollution (NO 2 and/or PM 10 ) were modelled and overall significant public health benefits were reported.
At the time of the study little epidemiological evidence of the impact of air pollution on health in Hungary was available. The health data used in this analysis was based on epidemiological research from other European countries which experiences much lower levels of air pollution. Hence the health benefits in these models may be uncertain.

Monetised-willingness to pay (WTP)
Health benefits from exposure to differing levels of air pollution (NO 2 and/or PM 10 ) were modelled: Excess mortality averted; Reduction in infant deaths; Annual symptom-days reduced; Annual respiratory symptoms days reduced.

Castro et al 2017; Switzerland
Health Impact Assessment Health outcomes: premature deaths, hospitalisation days due to CVD, RD; incident cases of bronchitis and asthma attacks in adults; cases of bronchitis and asthma symptom days in children; restricted activity days, working days lost.
Reduction of 3.3ug/m 3 PM 10 suggested prevention of 26 premature deaths, 100 hospitalisation days due to CVD, 110 days due to RD; 30 incident cases of bronchitis and 450 asthma attacks in adults; 150 prevalent cases of bronchitis and 1000 asthma symptom days in children; 47000 restricted activity days, including approx. 11000 working days lost.
Need for harmonizing HIA to allow direct comparisons between related or competing policy frameworks.
Modelling 2005 (counterfactual scenario) to 2015 (reference case) Assessments based on NO 2 reduction of 5.6ug/m 3 suggested prevention of 51 premature deaths.
The reduction in air pollution between 2005 and 2015 resulted in annual benefits valued at CHF 36 million (PM10) to CHF 49million (NO2).
Impacts of air pollution calculated using population attributable fractions (PAFs).
Monetisation of health impacts. Modelled reduction in PM10 and NO2 Sources were identified from World Health Organization meta-analyses • ReCiPe Overall, nuclear and renewable energy and natural gas power generate substantially less human health impacts than hard coal and lignite (fossil-fuels).
• IMPACT2002+ Fossil fuel combustion, mining (coal, uranium and metal) are the life cycle stages generating highest human health impacts.

Zvingilaite 2011; Denmark
Energy system modelling methodology paper that examines the inclusion of health externalities into the modelling to investigate optimisation of the model.

Health costs
Including health externalities into the planning of energy systems is more economical than paying for resulting damages later.
Important to include modelling of health externalities in planning energy transitions systems. Total health costs decrease approximately 18% and energy system costs reduce by nearly 4% when health externalities are included in the optimisation. Modelling two future scenarios for vehicle fuel use in Brazil: The population-weighted exposure to PM 2.5 and O 3 was 3.0 ug/m 3 Numerous assumptions could affect the results: vehicle fleet composition; prevalence of sugar cane burning; changing population demographics as an emerging economy. Biofuels may not be a solution to traffic related air pollution but combinations of improved vehicle technologies, economic incentives and shifts towards mass transit and active travel may be more important for public health. transport fuels (Tan et al 2016) and 'third-generation' liquid biofuels which include those produced from algae (Raheem et al 2015). To date, research is progressing into varying fuel stocks for biofuels but the 'secondgeneration' and 'third-generation' processes have yet to be shown to be commercially viable (Jose and Archanaa 2017).
Since the early 2000's there has been rapid global expansion in the biofuel industry-global biofuel output rose from 38 billion litres in 2005 to 131 billion litres in 2015 (Naylor and Higgins 2018). As it is an industry that is increasing exponentially, it is important that potential health effects are identified, assessed and mitigated. Potential health effects may arise via direct and indirect pathways. An in-depth review of the potential for health impacts of biofuels was conducted in late 2012 (Scovronick and Wilkinson 2014). This review found only five studies which were observational cross-sectional studies or HIA in nature (table 3). The linkages between biofuel production and use, and the pathways of exposure and health outcomes are multiple. The pathways of exposure may be through oral ingestion, inhalation or dermal contact with the fuel, with health effects varying depending on the chemical and the dosage.
There may be marked variability in risks to health associated with occupational exposures such as agricultural activities that can cause injury or disease, exposure to biological and chemical agents used in production and processing of the crops (e.g. herbicides, pesticides, ammonia, sulphuric acid, fungal spores, enzymes, antibiotics, ethanol), or exposure to biodiesel by-products, such as volatile organic compounds. There may also be risks to health through soil and water contamination from crop growing. Biofuel production requires much more water than fossil fuel production per unit of energy produced and, as such, expanded biofuel production may also contribute to local water shortages. The impact of these potential risks may vary depending on the geographic location, local ecology and site-specific legislation and practices (Scovronick and Wilkinson 2014).
A HIA of exposure to fossil fuel/petroleum versus ethanol and biodiesel in biofuel workers indicated that the biofuels emitted fewer carcinogens (Fink and Medved 2013). However, the biofuels emitted more organic respirable compounds, NOx and ionizing radiation than fossil fuels, and these could have potential health effects. The level of health impact identified in the HIA varied depending on the origin of the biofuel (Fink and Medved 2013). A cross-sectional study of workers in biofuel power plants compared to workers in oil and gas power plants reported that working in a biofuel plant did not seem to entail any greater additional risk for airway diseases compared with working in conventional energy plants (Schlunssen et al 2011). However, increased endotoxin and fungal spore exposure appeared to be associated with a higher risk of rhinitis (OR=3.1, 95%CI 1.1 to 8.8) and asthma symptoms (OR=8.1, 95%CI 1.5 to 44.4) among the biofuel workers (Schlunssen et al 2011). A cross-sectional study of the respiratory function of 39 wood pellet manufacturing workers found a significantly higher prevalence of self-reported nasal symptoms, self-reported breathlessness and asthma exacerbations. However, there was no significant difference in lung function among those who had worked longer in this setting or when compared to the selected controls (men working at a foundry) (Löfstedt et al 2017). These findings were limited by the small sample size and the cross-sectional study design as measurements were only taken at one point in time. Furthermore, the statistical methodology and results were not clearly reported. It is unclear if and how these findings might extend to general community exposures to biofuels.
Sugar cane is an identified source of energy for biofuel production. Burning of sugar cane straw is a common practice to enable easier access to the cane and to remove unwanted wildlife from cane fields. These burns are a major source of PM 2.5 during burning season which can persist for months. Epidemiological studies have reported associations between sugar cane burning and hospitalisations for asthma, hypertension and respiratory conditions among agricultural workers and communities exposed to the dispersed smoke (Scovronick and Wilkinson 2014). Clearly, these effects are similar to those experienced with general biomass burning, which have the potential for respiratory health impacts (Sigsgaard et al 2015). Of note, these impacts are likely to occur in and disproportionately affect communities more closely located to crop production and processing, compared with the general population where the fuels are ultimately used.
The beneficial effect, or otherwise, of switching from fossil fuels to biofuels in vehicles is not clear-cut. Studies of lower proportion biodiesel blends appear to show decreased emissions of PM 10 , hydrocarbons and carbon monoxide, but report increases in NOx (Scovronick and Wilkinson 2014). The research on air toxin emissions from biofuel production is unclear, with both increased and reduced emissions reported in the studies included in this review. Differing fuel blends may produce PM 2.5 and PM 10 with varying composition, size and structure which may lead to varying health risks related to their toxicity and oxidative stress responses (Betha andBalasubramanian 2013, Scovronick andWilkinson 2014). A simulated modelling study assessed the impacts of blending 7% and 20% of biodiesel to automotive diesel, in large cities in Brazil, on PM 2.5 emissions and subsequently on cardiorespiratory morbidity and mortality. The results indicated that 20% biodiesel blends were estimated to reduce morbidity and mortality, however they did not evaluate the potential health effects of NOx production and the secondary formation of ozone (O 3 ) (Vormittag et al 2018). Air pollution may also occur  No significant difference in lung function between the exposed workers and the controls; nor between workers who had undertaken the current tasks for less than 5 years or 5 years.
Authors imply that exposure to monoterpenes or dust during production of wood pellets was not an occupational risk to health.
Interpretation of results are limited due to methodological issues. Small sample assessed at one point in time only.
Peak exposures to dust and monoterpenes were not associated with acute effects on lung function.
Results of regression analysis were not reported. No adjustment for potential confounders or investigation of potential interactions.
No changes in nasal PEF between work and leisure time.
39 men (mean age=38 years (range 21-63 years) working in wood pellet production in six plants.

Health impact assessment
Fink and Medved 2013; No specific location; Health impact assessment (HIA) Biofuels (e.g. sugar beet bioethanol, soybean biodiesel, sugarcane bioethanol) are potential substitutes for fossil fuels in transportation.
Effect of biofuel production on workers: 1. Carcinogens 2. Respirable compounds 3. Ionizing radiation 4. UV-B radiation Outcome=DALYS Production of fossil fuel/petrol emits more carcinogens than sugar beet ethanol, sugar cane and rapeseed biodiesel. Higher health impacts from organic respirable compounds emitted during biofuel production compared to fossil fuels. Sugar beet ethanol and soybean biodiesel affects human health less with inorganic respirable compounds HIA of selected firstgeneration biofuels shows some advantages with regards to less carcinogenic compounds and nonionizing radiation. Majority of health effects in production of liquid biofuels comes from organic and inorganic respirable compounds, but level of effect Modelling methodology not clear. This study did not evaluate the potential health impacts of other secondary air pollutants associated with biodiesel such as NOx, ozone on health.
Increasing to B20 over the study period-estimated 13,031 fewer deaths and 28,170 fewer hospitalisations. Project funded by a biodiesel producers' association.
Simulation modelling to estimate impact of addition of biodiesel to diesel for automotive use over the period 2011 to 2025. Scenarios=7% (B7) and 20% (B20) at other stages of the biofuel life-cycle, and the benefits and adverse impacts may be differentially experienced across geographic regions, suggesting potential spatial variation in health impact.
These findings indicate the scarcity of health data related to biofuel production, handling, use, disposal and variation by location. This highlights the need for HIAs to include the LCA of the range of biofuels in order to better understand the potential for health impacts, both adverse and beneficial.

Wind energy
The most health-related research in the energy transitions/renewable energy field was evident for wind turbine and wind farm operations. The driving force for this research has been community concern over the alleged health effects experienced by some people living near wind farms. The key aspects of wind farms that are complained about are noise, shadow flicker from turbine blades, electromagnetic radiation and infrasound (inaudible sound). Alleged health effects that have been investigated include sleep disturbance, insomnia, headache, tinnitus, nausea, tachycardia, problems with concentration and memory, panic episodes, and photoinduced epilepsy (Knopper andOllson 2011, Jeffery andKrogh 2014).
Systematic reviews conducted up to 2015, of the health impacts of exposure to wind turbines/farms, found that there is no consistent evidence that wind farms cause adverse health effects. However, the reviews concluded that higher quality studies are warranted, especially for those people living close to wind farms (i.e.  4). The overall findings from these papers did not shed alternative findings to the systematic reviews reported previously. The evidence over whether wind turbine farms are associated with negative health effects is still hotly contested. Overall, this most recent research has indicated that stronger adverse health effects were associated with negative attitudes towards wind turbines including concerns regarding property devaluation, visual impacts and noise sensitivity. Two government funded studies on the health impacts associated with exposure to wind turbine noise or infrasound are currently underway in Australia (https://windfarmstudy. com/?/home [Accessed 25 September 2019]) These include both laboratory-based control studies and field studies of controlled exposures. They are due to report their findings in 2021-22.

Photovoltaic cells (solar panels)
In simplified terms, solar power is a form of renewable energy that is produced via photovoltaic (PV) cells which absorb photons from the sun's rays to excite the electrons in the PV cells resulting in electricity production. This electricity can then be used to supply renewable energy as single semi-conductor cells (e.g. solar powered calculators, and watches) or assembled and encapsulated into solar panels. Solar panel technology is improving, and the technology is becoming increasingly accessible to populations across the world. This electricity production results in reduced gas and particulate emissions compared to electricity produced from fossil-fuels, such as coal (Abel et al 2018a). A number of studies have modelled GHG emissions and air pollution levels, and extrapolated that decreased emissions were likely to lead to reduced health impacts (Siler-Evans et al 2013, Wiser et al 2016, Abel et al 2018b. Despite the rapid improvements being made to PV cells and the uptake in use for electricity production, we found few papers that specifically examined the impacts of PV cells/panels on human health. The life cycle of solar PV panels, incorporating their production to end-of-life, raises potential health and environmental issues. The structure and design of PV cells, panels and modules vary depending on their application. In general, there are four broad families of PV cells/modules (ranked from most expensive and efficient to the least efficient): (1) mono-crystalline silicon-single silicon crystal cut into wafers approximately 0.2mm thick; (2) poly-crystalline/multi-crystalline silicon-cells containing many small silicon crystals; (3) thin film-crystalline cells cut into wafers of 2μm thick (layers of this film containing amorphous silicon, cadmium telluride (CdTe), copper indium selenide (CIS) or copper gallium selenide (CIGS) are placed on glass forming a panel similar to polycrystalline modules; these use less material and are cheaper but are also less efficient), and; (4) multi-junction panels comprised of indium gallium phosphide (InGaP), gallium arsenide (GaAs) or indium gallium arsenide and germanium cells (InGaAsGe) (Bakhiyi et al 2014). The production of PV cells involves exposure to a range of heavy metals, chemicals, acids, bases, gases and solvents (for example: aluminium, arsenic, asbestos, cadmium, carbon tetrachloride, copper, hexavalent chromium, hydrofluoric acid, lead, ammonia, argon gas, hydrochloric acid, methane, silane gas, tellurium and nitrogen trifluoride), which may have non-carcinogenic and carcinogenic health effects (Aman et al 2015). Silver is used in PV cell manufacture and is considered a relatively valuable metal (Kuczyńska-Łażewska et al 2018), and so there is the risk that increased PV cell production to meet rising global demand will place undue pressure on existing silver resources. Participants: opponents, wind farm hosts and 'fence-sitters'.
The 'facts' about whether wind farms cause negative health effects are contested.
Found that stake in windfarm, interest and legitimacy are particularly relevant for the competing descriptions about the 'facts' of wind turbine health effects.
Purposively selected participants. Discursive psychological assessment of how people talk about the health effects of wind farms -conversation analysis.
No objective health measures. Questionnaire: response to the environment, perception of wind turbine noise, implementation of sound mitigation measures on houses.
Key findings: exposure to wind turbine sounds significantly impairs individual wellbeing via the strong effect it has on their decision to spend resources in retrofitting their houses to minimise perceived sound. This is independent of reported annoyance.
More objective data needed to assess the impact of wind turbine noise on individual health or well-being.

No objective health measures. Portugal
Compensation may be needed to allay retrofitting costs.

Noise measurements
Questionnaire. Response rate=38% Economic compensation did not appear to act as an effect modifier.

Response bias
Norway.
Annoyance rather than health effects examined.
Noise annoyance depends strongly on separate non-acoustic factors: visual and aesthetic factors. Cross-sectional survey Socio-acoustic study post installation. Wind farm (31 turbines) that affects 179 dwellings within 2km radius (n=90).

Kageyama et al 2016
Noise measurements in seven locations within 1km of nearest wind turbine, excluding road traffic noise.
No association between noise exposure levels with poor physical/mental health was found.
Sensitivity to environmental stimuli should be considered in future field studies.
Wide confidence intervals. Japan Socio-acoustic study Cross-sectional study. Rural areas.
Significant association between outdoor wind turbine noise exposure and self-reported insomnia (  Small sample is a major limitation.
Ontario, Canada Participants reported poorer sleep quality if they had negative attitudes to wind turbines, concerns regarding property devaluation, and visual impacts. Self-reported sleep may be associated with indirect effects of visual and attitudinal cues or concerns regarding property devaluation.
Prospective cohort established before installation of wind turbines.
Associations between noise exposure and sleep parameters were not calculated as the number of participants was too small (n=3).
Before and after installation of wind turbines.

Polysomnography to assess sleep quality
Results from polysomnography showed that sleep parameters were not significantly changed after exposure. However, reported sleep qualities were significantly (p=0.008) worsened after exposure. Noise levels in participants' bedroom did not change between before and after wind turbine installation.
This study cautiously suggests that there are no major changes in the sleep of participants who live near new industrial wind turbines in their community.
Lack of control group, with regard to the exposure levels and wind speed, and with other possible sources of variation that might affect results.
Prospective cohort of 16 adults living within 1 km and in view of a wind turbine.
Assessed sleep parameters at two time points. Wind turbine noise levels were not found to be related to scores on the Physical, Psychological, Social or Environment domains, or to rated QOL and Satisfaction with Health questions.
Results do not support an association between wind turbine noise levels and decreased QoL using the WHOQOL tool.
Reporting bias.
Hearing wind turbines for less than one year (compared to not at all and greater than one year) was associated with improved scores on the Psychological domain (p=0.01). Lower scores on both the Physical and Environment domains (p=0.02 and p=0.04),were observed among participants reporting high visual annoyance toward wind turbines. Personal benefit from having wind turbines in the area was related to higher scores on the Physical domain (p=0.04).
communities with wind turbine farms.
The electricity generated by the PV cells/panels needs to be initially stored in a battery group so that it can be supplied as needed, hence rapid expansion of the PV system requires expansion of battery production and disposal. These batteries contain lead and acid which, if not managed properly, can adversely impact on the environment and human health. Some gaps in LCAs of PV panels have been identified. For example, during the production phase the quantification of emissions of fluorinated-gases and other by-products needs to be undertaken, and reporting of data on specific air emissions and liquid/solid effluents needs to be improved. During the PV operational phase there is uncertainty over: toxic emissions in the event of a fire; the level of potential for toxic rainwater to leach into home water supplies, stormwater or land surface run-off; the longevity of the solar panels; and the risks to PV cells during extreme weather events. During end-of-life processing, the toxic potential of PV cell waste in landfill or incineration needs to be quantified in relation to potential contribution to soil contamination and air pollution. Other considerations during this phase include the impacts of decommissioning, dismantling, and transporting the PV panels for disposal and the associated electricity demand (Aman et al 2015). Similarly, there is a need to examine the potential health and environmental impacts of batteries and the prospects for recycling of the batteries (Xu et al 2018).
These findings indicate the need for life-cycle HIA of PV systems to better understand potential health and environmental impacts, both adverse and beneficial, across the production, operation, end-of-life, disposal (including take-back) and recycling of PV cells and batteries (Xu et al 2018). Such HIAs need to determine the likelihood and magnitude of risk to enable appropriate risk management procedures to be implemented across the industry.
Electric and hydrogen fuel-cell vehicles As of 2017 there were more than 2 million electric vehicles in service globally with electric vehicles representing an increasing proportion of new car sales (Wilberforce et al 2017, Requia et al 2018. Electric vehicles are regarded as a key technological development to support sustainable transportation and mitigate the impacts of climate change through reduced GHG emissions. An expected co-benefit of reduced traffic-related air pollution from electric vehicles compared to internal combustion engines using fossil fuels or biofuels is improved public health outcomes (Navas-Anguita et al 2018) (table 5).
However, these environmental and health co-benefits will only be realised if the source of electricity used to power the electric vehicles derives from low/no carbon renewable energy sources (Jacobson et al 2005). Where the infrastructure used to power electric vehicles relies on conventional fossil fuel combustion, e.g. coal based power stations, then inequity of benefits can occur when there is an unequal burden of polluting by-products in areas where benefits of electrified vehicles are not experienced (Ji et al 2015). For example, electricity generating plants may be located in areas where populations are less likely to be able to afford or use electric vehicles. These areas are at greater risk of being exposed to higher levels of air pollutants. Technically, electric vehicles may have net benefits if charged with gas-or renewable energy-powered electricity and those power plants are located far away from people. With increasing use of electric vehicles, we need to consider the location and sources of electricity production and emissions produced, in order to maximise distributional fairness of impacts. Transitioning from passenger vehicles to active transport (walking, cycling) and reducing the numbers of vehicles on the road have been shown to have beneficial health impacts associated with reduced air pollution, increased physical activity, and reduced environmental noise (Perez et al 2015, Xia et al 2015 (table 5).
Although tailpipe emissions from fossil-fuelled internal combustion engines will be reduced in electric vehicles, other emissions such as particulate matter from tyre and brake wear and roadway dust dispersion remain, and these have the potential to impact on health.
Much research and development is being undertaken to design cost-effective electric car rechargeable batteries to store more energy and lengthen the distances and travelling times (Grey and Tarascon 2016). This form of technology offers great potential for electrification of mass transport systems (Borén et al 2017). As with solar panels there is a need to investigate the life-cycle HIAs of battery use.

Building energy efficiency
The aims of improving residential energy efficiency stem from the desire to reduce energy consumption, reduce the demand for fossil-fuels, alleviate financial hardship on households and reduce thermal impacts on health. Several review papers have examined the complex relationship between improving residential energy efficiency and health outcomes (Maidment et al 2014, Willand et al 2015, Willand et al 2017. The papers included in this review are grouped into study type and summarised in table 6. A meta-analysis of 33 building energy intervention studies (installing insulation, central heating, double glazing of windows) that included approximately 33,000 resident participants found that, on average, the interventions led to small but significant improvements in self-or parent-reported health status (Maidment et al 2014). However, only four studies collected objective measures of health outcomes, for example, lung function tests, blood tests, medical examination, or blood pressure. Overall, it appears that programs that addressed known problems, such as dampness, cold, or insulation from cold or heat, had more impact than those that addressed broader energy efficiency aspects (e.g. the desire to reduce energy consumption overall). Positive health effects were reported in studies of children; studies of children, adults and older people with poorer health status; and studies of people living on low-incomes. Larger health effects were seen in urban areas however this effect may be biased by the use of objective health testing in these settings or the increased exposure to outdoor air pollution which the interventions provided some protection from.
Other reviews of residential energy efficiency interventions explored the contextual influence on health outcomes (Willand et al 2015, Willand et al 2017. The key messages from these reviews were that residents' expectations influenced their overall satisfaction with the interventions. In addition, cultural practices around heating of homes such as providing excessive ventilation, resulted in reduced indoor temperatures, despite the attempt to improve indoor warmth. Furthermore, economic deprivation and mastery of technology continued to impede acceptance of energy efficient interventions and energy efficiency. A number of multi-disciplinary housing studies reported that working in partnership with communities and government agencies to retrofit insulation and install more effective heating has led to significant improvements in health and wellbeing, especially in low-income housing of vulnerable people (Breysse et al 2011, Howden-Chapman et al 2011, Garland et al 2013, Grey et al 2017. Some studies also suggested that improving energy efficiency in the home, by reducing air leakage and airflow, may have deleterious health effects because of increased potential for growth of microorganisms such as mould, fungi, house dust mites and bacteria. It is recommended that ventilation measures for health protection and the potential variation in the impact of home energy efficiency strategies be considered in the intervention design of any household energy efficiency program (Gens et al 2014). Importantly, research has shown that there is a need for tailored policy approaches in different locations and climates, rather than simply adopting universally rolled out strategies (Shrubsole et al 2015).

Implications and conclusions
The field of energy transitions is broad, complex and developing rapidly as governments and industries globally move to adopt policies and targets to achieve a reduction in carbon emissions. We consider this scoping review to be a first step in highlighting potential health impacts of specific energy transition processes and technologies that might otherwise not be fully explored in the literature from a public health impact perspective. Our literature search indicated that, to date, it appears that the depth and breadth of the health impact research is very limited, especially in comparison to research on climate or energy return on investment. It is possible that our search did not produce all relevant papers as we did not include specific health-related search terms such as mortality, morbidity and cancer, amongst others. However, given that the search identified 6933 abstracts for screening, including abstracts with these terms, we are of the opinion that, in all likelihood, our review was successful in identifying the majority of relevant papers.
Research that examines health impacts of energy transitions needs to be multidisciplinary and continually evolving to keep up with the technological developments and policy shifts. From a public health perspective, we strongly support measures to facilitate the transitioning of carbon-based energy use to lower and non-carbon energy sources, as there are quantified health benefits of reduced airborne pollutant emissions from this transition. However, we also acknowledge the need to determine the potential for unintended adverse health impacts arising from the adoption of new measures and technologies.
Our search terms captured a broad range of literature related to energy transitioning, but the depth of research identified and reviewed was limited in some areas given that the search focussed on the health impacts of energy transitions. To better understand the depth of research in each energy transition area, additional individual systematic reviews would need to be undertaken. However, in-depth reviews on each energy theme were beyond the scope of this review. This review did however identify up-to-date in-depth literature reviews which informed some of our findings.
Epidemiological studies examining the health effects associated with a range of energy transition forms were scant. This is perhaps not surprising, given the difficulty in conducting well designed epidemiological studies within this domain. Health impacts were most commonly derived from modelling studies that utilised existing prevalence data for a range of health conditions which were expected to be affected by the environmental exposure/s being examined. The key modelling studies that analysed health co-benefits examined changes in air pollution levels associated with climate change policies, increasing energy demands, and altered vehicle emissions.
We anticipate that this review might subsequently lead to the need for more targeted research to fully explore the impacts on health arising from specific technological or policy changes related to transitioning between Scenarios of electric vehicle (EV) penetration into the Spanish transport sector over 30 years (2020,2030,2040,2050).
Human health impacts -DALYs Coal-fired power plants are the most damaging power generation technology in terms of health, so partial avoidance gives rise to favourable reduction in DALYs. However, the withdrawal of fossil-based power generation (natural gas, cogeneration) have less significant impact. There was an overall trend in increasing DALYs for all 3 scenarios after then avoidance of coal-powered generation is no longer happening.
Increased electricity demand in Spain likely to be met by onshore and offshore wind power-this would lead to slight increase in annual life cycle impacts of the power generation sector. High market penetration of 20 million EVs by 2050 could be 0.25 DALYs. This minor impact is likely to be offset by high environmental benefits due to the avoidance of fossil fuel use in the transport sectors-predicted net annual savings of 4-9 DALYs. This modelling suggested that currently planned approaches 'DP' will bring relatively large air pollution health benefits, principally due to reduction in tail pipe emissions. Four modelling scenarios: Considerably less change (<2%) in Lden and Lnight for any scenarios considered.
The more ambitious hypothesized scenarios considering large penetration of electric cars in the city in the year 2020 did not contribute considerably to increased health benefits from noise reduction and that an increase in population exposure to noise and related negative health impacts is even predicted under the DP scenarios. Despite moderate benefits of air pollution reduction, this 1. 'Decided policies' (DP) DP: 3% (65) reduction in natural deaths. study indicates that noise reduction has the largest health effectiveness ratio when the energy production is principally from renewable energy.
Limitations: uncertainties are not quantified, not all assumptions are validated. Modelled a range of scenarios. The largest health benefits would occur when increased public transport and cycling are combined, which is estimated to result in a 55% reduction of total disease burden attributed to physical inactivity.

Health impact assessment
Comparative Risk Assessment Approach (5% and 10% reduction in passenger vehicles). The health impacts calculated as population attributable fractions (PAFs) for shortterm and long-term PM 2.5 exposures were estimated.
All models resulted in reduced PM2.5 and CO2 emissions. Exposure=PM2.5 PAFs for short-term and long-term PM2.5 exposures were estimated to decrease, in line with reduced PM2.5. The total burden of disease prevented from air pollution reduction was estimated to be 39 DALYs in both 'Increased Cycling scenarios', and varied from 52 to 98 DALYs in the 'Increased Public Transport' scenarios. The most substantial health benefits came from the reductions in disease burden associated with ischaemic heart disease and stroke.  Domestic energy efficiency program.
Self-reported physical and mental health outcomes using the SF-12v2 composite scales and subjective well-being.
The energy efficiency programme was not associated with improvements in physical and mental health or reductions in selfreported respiratory and asthma symptoms. However, the programme was associated with improved subjective wellbeing (β=0.38, 95% CI 0.12 to 0.65), as well as improvements in a number of psychosocial outcomes, including increased thermal satisfaction (OR=3.83, 95% CI 2.40 to 5.90), reduced reports of putting up with feeling cold to save heating costs (OR=0.49, CI=0.25 to 0.94), fewer financial difficulties (β=−0.15, 95% CI −0.25 to −0.05), and reduced social isolation (OR=0.32, 95% CI 0.13 to 0.77).
Investing in energy efficiency in lowincome communities does not lead to selfreported health improvements in the short term. However, investments increased subjective wellbeing and were linked to a number of psychosocial intermediaries that are conducive to better health. It is likely that better living conditions contribute to improvements in health outcomes in the longer term.
Relatively large sample.
Quasi-experimental field study Self-reported respiratory and asthma symptoms.

Short term impacts
Pre-test-post-test Potential contamination of control group.
Self-report: Respiratory symptoms and medication usage.
Indoor temperatures increased by 1.1 degree Celsius in living room and 0.53 degree in child's bedroom.
Multidisciplinary housing studies show that working in partnership with communities and government agencies to retrofit insulation and install more effective heating has led to significant improvements in health and wellbeing.
Response bias may impact results.
New Zealand Indoor temperature.
Self-administered lung function tests: peak expiratory flow rate (PEFR) and forced expiratory volume in 1 s (FEV1).

Levels of NO 2 halved.
Did not report on potential contamination of the intervention/control sites. Community-based randomised controlled trial Levels of NO 2 .
Data linkage: general practitioner visits and hospitalisations.
Parents in the intervention group reported less poor health (OR=0.44, 0.28-0.7).
Limited discussion of the impact of moderating or interacting characteristics.
N=409 household with a child with doctor diagnosed asthma.
Sleep disturbance due to asthma symptoms reduced significantly.
No difference in lung function between intervention and control group.

Modelling
Gens et al 2014 Impact of improved insulation on indoor particulate matter.
Compared to 0% insulated, modelled scenarios of 50% and 100% led to increased DALYs. Opposite contributions identified: (1) The reduction in outdoor PM emissions due to reduced energy demand results in decrease in DALYs: CH & GR=2500; CZ=5000.
Both effects together indicate that accumulation of PM indoors if high indoor PM sources are present. The effect of these PM accumulations may outweigh the benefits of reduction in outdoor PM on the population average.
Only PM considered, not impacts on levels of fungal spores, radon or relative humidity and associated health effects. Switzerland (CH), Czech Republic (CZ) and Greece (GR).
Need to consider ventilation when increasing energy efficiency of buildings. To examine changes, 2010-2050, in end-use energy demand, CO 2 emissions, winter indoor temperatures, airborne pollutant concentrations and associated health impacts: all cause and cardiovascular, cerebrovascular, myocardial infarction, cardiopulmonary, lung cancer mortality data.
The average net impact on health (change to life-expectancy at birth) per 1000 population was greater in magnitude under all scenarios in London compared to Milton Keynes and more beneficial when it was assumed PPV would be part of energy efficiency interventions (London ∼+4 months; MK ∼+3 months) , but more detrimental when interventions were assumed not to include PPV (London~−5 months; MK ∼−2 months). Important to consider ventilation measures for health protection (not adversely affecting indoor air quality) and the potential variation in the impact of home energy efficiency strategies, suggesting the need for tailored policy approaches in different locations, rather than adopting a universally rolled out strategy.
Modelling relies on assumptions and hold many uncertainties. Results are indicative and relative, rather than evidence of direct impact.

31
Environ. Res. Commun. 2 (2020) 065003 energy sources, for example, waste to energy technologies or electrification of vehicles. We recommend that the consistent gap in knowledge that has emerged from this review could be addressed by conducting life-cycle HIA or modelling studies of current and developing energy transitions technologies, interventions, and policy decisions on health. There is a need to conduct individual systematic or in-depth reviews for each of the energy transitions themes to identify the key stages that would inform the development and implementation of lifecycle HIAs or modelling. Toxicological research could also inform the development of life-cycle HIAs for this purpose. An example of an energy transition field that would benefit from toxicological data is that related to biofuel production and use. Given the rapid speed with which some of these energy transitions are occurring it is imperative that such assessments and studies be conducted as soon as possible so that policy decisions and investment priorities are supported by a solid evidence base that protects not only the environment but also public health.

Funding
This review was funded through a seed grant from the National Health and Medical Research Council funded Centre of Research Excellence, Centre for Air pollution, energy and health Research (CAR)(NHMRC APP1030259; [SEED05.2017]).